Messages in this thread | | | From | (Eric W. Biederman) | Subject | Re: [PATCH] fs, proc: Introduce /proc/<pid>/task/<tid>/children entry v7 | Date | Thu, 19 Jan 2012 15:14:05 -0800 |
| |
Cyrill Gorcunov <gorcunov@gmail.com> writes:
> On Thu, Jan 19, 2012 at 06:57:00PM +0100, Oleg Nesterov wrote: >> On 01/19, Cyrill Gorcunov wrote: >> > >> > On Thu, Jan 19, 2012 at 07:51:12PM +0400, Cyrill Gorcunov wrote: >> > > If it's needed I can wrap all this with CONFIG_CHECKPOINT_RESTORE, should I? >> > > >> > > --- >> > >> > Oleg, if only I'm not missing something obvious you meant handling like below? >> >> Yes, but... >> >> > +struct proc_pid_children_iter { >> > + struct pid_namespace *pid_ns; >> > + struct pid *parent_pid; >> > +}; >> >> you forgot to remove this definition. >> > > No, I rather forgot to quilt refresh :) > >> > +static int children_seq_show(struct seq_file *seq, void *v) >> > +{ >> > + struct inode *inode = seq->private; >> > + unsigned long pid; >> > + >> > + pid = (unsigned long)pid_nr_ns(v, inode->i_sb->s_fs_info); >> > + return seq_printf(seq, " %lu", pid); >> > +} >> >> just noticed... why unsigned long and %lu? afaics pid_t/%d should work >> without any typecasts. >> > > I'm not sure how important it is, but Andrew mentioned in one of email > that we might be moving from pid_t from int to long one day (which of > course will require extreme huge work on checking code where int->long > transition might cause problems). So I thought why should I wait then? > > [ Andrew, am I correct? ]
It is going to take a lot to get to a > 32bit pid value. I would not worry about it today unless there is some 64bit arch somewhere that defines pid_t as long.
Eric
| |